Bengaluru-based AI startup Sarvam is making a strategic pivot, expanding its ambition beyond merely developing AI models to establish itself as India's premier full-stack sovereign AI platform. This move positions Sarvam to directly challenge global AI leaders, including major model developers and hyperscale cloud providers, by offering a comprehensive ecosystem of infrastructure, foundation models, enterprise agents, and developer tools.
A Shift Towards Integrated AI Solutions
At its inaugural EPOCH conference, Sarvam outlined an ambitious roadmap, signaling a belief that the next wave of AI adoption in India will be driven by complete AI stacks that enterprises can deploy within the country, rather than by isolated models. Founders Pratyush Kumar and Vivek Raghavan emphasized their vision for Sarvam as a horizontal platform where diverse applications can be built.
Expanding Product Portfolio and Sovereign Stack
Sarvam unveiled a suite of new products designed to serve various enterprise needs. These include advanced voice agents, coding assistants, enterprise work agents, sophisticated speech recognition and text-to-speech models, document intelligence solutions, and AI-powered dubbing services. The company also introduced a 'model training as a service,' enabling enterprises to fine-tune models on their proprietary datasets with minimal engineering effort.
Central to Sarvam's strategy is its concept of a sovereign AI stack: AI models developed from the ground up, hosted entirely within India. This approach prioritizes data residency and regulatory compliance, crucial factors for Indian enterprises. Sarvam claims its voice and chat models are significantly more cost-effective—five to ten times cheaper—than comparable frontier offerings, all while ensuring customer data remains within Indian borders.
Major Investment in AI Infrastructure
Perhaps the most significant announcement concerns infrastructure. Sarvam disclosed its current access to approximately 2,000 NVIDIA Blackwell GPUs and articulated an aggressive goal to scale this to 10,000 GPUs. These investments extend to building India-based AI infrastructure and sovereign data centers, highlighted by a strategic partnership with HCLTech in Odisha. The startup also committed to developing a trillion-parameter foundation model within the next six months, underscoring its belief that India must cultivate its own frontier AI capabilities rather than relying solely on international providers.
Becoming India's Domestic AI Partner
Sarvam is not primarily positioning itself against global giants like OpenAI. Instead, it aims to be the domestic partner of choice for Indian enterprises seeking an integrated AI solution that encompasses infrastructure, models, deployment, and ongoing support. The company's founders stated their objective to serve a substantial portion of the AI tokens consumed across India, viewing infrastructure, model training, and enterprise applications as components of a unified platform strategy.
While specific financial details remain undisclosed, Sarvam reported that over 500 enterprises, startups, and organizations are currently leveraging its products, with the majority of its revenue stemming from the private sector. This strategic pivot reflects a broader evolution within India's AI ecosystem, where competitive advantage increasingly lies in owning the entire AI stack, from compute infrastructure to enterprise applications and sovereign deployment.